- Published on
Databricks co-founder shares his secrets for winning big enterprise customers
- Authors
- Name
- DeepHorus
At TechCrunch Disrupt last week, Databricks co-founder and senior vice president of field engineering Arsalan Tavakoli-Shiraji delivered a blunt message to startups chasing large enterprise contracts: your technology alone will not win the deal. Drawing on his experience landing Databricks' first major customers, Tavakoli-Shiraji warned that founders who believe superior tech sells itself are making a critical and costly mistake. His advice arrives as startups face mounting pressure to grow quickly and tap into enterprise IT budgets, making the playbook for winning big customers more relevant than ever.
Databricks, now valued at $190 billion, counts more than 70% of the Fortune 500 as customers and has become a giant in AI and big data storage. But Tavakoli-Shiraji recalled the company's early days when it was a tiny startup struggling to get meetings with major prospects. One early meeting collapsed because the prospect had already bet his career on a Databricks competitor, while other deals fizzled due to unpreparedness on legal, contractual, and security questions. Those hard lessons taught him that "people and processes matter the most" in enterprise sales, not just technical superiority.
His first tip is to sell under the umbrella of a larger technology company. When Databricks failed to secure a meeting with UBS, it partnered with Microsoft as an Azure cloud partner, allowing sales agreements to be covered under Microsoft's terms. Microsoft's sales team then booked meetings and advocated for the startup, dramatically improving access and distribution. Today, Databricks plays the role of big-company umbrella for startups like Hightouch, Addepar, and Tecton, selling their products under its own terms and cybersecurity setup, demonstrating how the strategy scales over time.
For AI products specifically, Tavakoli-Shiraji advises addressing what he calls the "four Cs": context, choice, cost, and control. Context means AI that can be trained like a new employee, learning company terminology, processes, and priorities. Choice is the freedom to switch models rather than being locked into one provider. Cost involves controlling how much the AI product costs to run, while control covers safeguards governing AI behavior and protecting sensitive information. These four concerns, he argues, are the real objections enterprises raise when evaluating AI solutions.
Tavakoli-Shiraji also urges startups to attack the expensive legacy systems that enterprises believe they are stuck with. He jokingly asks CIOs to rate their AI adoption on a scale of one to five, where one means employees merely have access to ChatGPT and five means AI has changed core business processes. Most executives claim to be a three leaning toward four, but he said 99% of those he talks to are actually in buckets one and two. Enterprises still have not figured out how to deploy AI agents to create value beyond helping engineering teams write code.
The implications for startups are significant: winning enterprise customers requires patience, partnership, and preparation for non-technical hurdles. Short-term, founders should prioritize channel partnerships and security readiness over product features alone. Long-term, the companies that succeed will be those that treat enterprise sales as a people-and-process discipline, not a technology showcase. Tavakoli-Shiraji's own journey from struggling startup to $190 billion giant suggests the approach can pay off, but only for those willing to learn the hard lessons early.
Industry reactions to his talk highlighted a familiar tension: investors and founders often over-index on technical differentiation, while enterprise buyers care deeply about risk, integration, and support. Tavakoli-Shiraji's remarks resonated with operators who have seen promising startups stall at the pilot stage because they underestimated procurement and compliance. As AI adoption accelerates, the ability to navigate the four Cs and leverage larger partners may become a more reliable predictor of enterprise success than benchmark scores or product demos.
Looking ahead, Tavakoli-Shiraji expects the deployment of AI agents in enterprises to take many years, even as excitement builds. Startups that align with Microsoft, Databricks, or other platform giants will likely gain faster access to enterprise buyers, while those that go alone will face longer sales cycles. The next wave of successful enterprise startups may not have the best tech, but they will have the best processes, partnerships, and understanding of what large customers actually need before they sign.
